基于联想记忆神经网络的控制方法

提出了一种新型联想记忆神经网络结构和学习算法，通过反馈通道将网络输出反馈至输入层，并在通道中引入联想记忆因子，提高了X寸非线性系统的辨识能力，能够简单有效的建立起电阻炉的数学模型。采用具有逆辨识结构的神经网络控制方法对电阻炉进行实际控制，逆控制器的参数通过神经网络逆辨识模型得到，神经网络采用3层2—8—1结构形式。与常规PID控制相比较表明，此方法对大滞后非线性系统具有良好的控制效果和较强的抗干扰能力。

Control Method Research Based on Neural Network

Aim at the problems of identification and control about time - delaying non - linear system such as resistance heating furnace, a kind of associative memory neural network structure and learning algorithm are pointed out. The output of network is feedback to input layer by feedback channel, associative memory factors is applied at the feedback channel. The identification ability to non - linear system is improved, and the mathematical model of resistance furnace could be set up easily and effectively. The neural network control method with inverse identification structure is adopted to control the resistance furnace in actual. The parameter of Inverse controller is received by neural network inverse identification structure. The neural network adopts 2 - 8 - 1 three lays structure. The result which comparing to the conventional PID shows that this method has a better control effect and anti -interference ability to time -dela- ying non -linear system.